Simon Maxwell

Keep in touch!

Re-thinking the concept of fragile states

Re-thinking the concept of fragile states




(Note: a pdf version of this piece is here, with slightly more readable graphics.)

The argument

This is the story of how I was intrigued by the different definitions of ‘fragile states’. I wanted to understand how we could be talking in the same breath about war-torn countries, like, say, Yemen; and climate vulnerable countries, like, say, the Marshall Islands; and countries with weak institutions, like, say, Burundi. I quickly became frustrated, and finally reached some radical conclusions:

First, that there are too many lists, with too many different definitions, and insufficient country overlap. Furthermore, some of the definitions are too wide, and some of the lists are too long. There are lines of causality linking different kinds of fragility, of course, but a conversation which puts all into the same pot loses focus. The debate risks sinking into a morass of generalisations.

Second, that the main challenge to traditional ways the aid industry does business is in conflict countries, and that it would be no bad thing to restrict the class of fragile states to those currently or likely to be conflict-affected, and possibly those just recovering from conflict. In these cases, there are genuine dilemmas about: tackling the drivers of conflict; the interface between development and humanitarian actors and warring parties; the role of peace-making and peace-keeping forces; and the transition from war to peace.

Third, that even here there are multiple and competing criteria, with different lists and different country coverage. The whole business of drawing up top down, data-driven categories and lists is fraught with difficulty.

Fourth, a better approach would be to use expert panels and participatory methods to develop flexible, up-to-date and relevant targets for analysis and intervention.

Fifth, the over-burdened concept of fragile states has become debased; and the concept of conflict-affected states should be restricted to the egregious cases.

And, sixth, this would allow everyone to focus bottom up on the dilemmas listed above, wherever they occur.

Reviewing the approaches

I began by assembling the lists from the World Bank, DFID, the New Deal, and the recent OECD/DAC States of Fragility Report. There are others, as we shall see. Interestingly, the Cameron Commission on fragile states, which reported earlier this year, eschewed lists altogether. In the four cases cited, there are 80 countries listed (plus, in DFID’s case, three whole regions): that is well over half the total number of countries on the OECD/DAC aid recipients list (currently 143). Only 13 countries appear on all four lists, and only a further 15 on three of the four. See the Venn diagram in Figure 1 (for country abbreviations, see Appendix 1).

Figure 1

Venn diagram of fragile and conflict-affected states

Sources: see text

For reference, the table in Appendix 1 shows which countries appear on which list – and this table is the source for the Venn diagram in Figure 1.

As noted, the Cameron Commission did not have a list.

The New Deal is a voluntary agreement involving fragile and conflict-affected states, to which countries sign up. According to the website, there are 18 developing country members.

The World Bank ‘Harmonized List of Fragile Situations’ is based on CPIA scores (CPIA being the Country Policy and Institutional Assessment), also including countries with UN or regional peace-keeping or peace-building missions. It includes data from the African and Asian Development Banks.  There is more detail in the footnotes to the table in Appendix 2.

The DFIDDepartment for International Development list is taken from the 2016 Bilateral Development Review, with the key information found in the background Technical Note. The relevant page is reproduced in Appendix 3. Fragile States are divided into three categories (high, medium and low), with an additional list of eligible countries which neighbour high fragility states. This gives a total list of 64, plus three regions designated as ‘fragile’, namely the Middle East, South of Sahara and North of Sahara. The Note cites data sources, but does not spell out the methodology. It also says that the list will be updated regularly: I have no information on whether this has yet been done. Independently of DFID, the UK Foreign Office runs the Conflict, Stability and Security Fund, which has operations in 70 countries, some of which (e.g. Morocco) are not on the DFIDDepartment for International Development list. See Appendix 4.

The DACDevelopment Assistance Committee (of the OECD) list comes from the State of Fragility Report, first published with this title in 2015, and designed to provide a more comprehensive view than previously attempted of political, societal, economic, environmental and security-related fragility. Some 60 countries were scored on each of these five dimensions, giving a multi-dimensional picture of fragility. The data were updated in 2018, giving the picture in Figure 2. The methodology is summarised in the 2016 report (which incidentally was focused on violence): Pgs 74 ff and Appendices from Pg 152. Inter alia, one learns that quite a few countries were missed out for lack of data (Figure 3): some of these appear on other lists of fragile states. In a key feature, the DACDevelopment Assistance Committee (of the OECD) scores each country on each of the five dimensions of fragility, allowing for disaggregated analysis. The 2018 scores are given in Appendix 5.

Figure 2

OECD Fragility Framework 2018


Figure 3


In sum, there is a great deal of heterogeniety. Sebastian Zaja and Javier Fabra Mata review the problems. They note that the criteria vary and that data are hard to acquire. When it comes to constructing indices, they identify other methodological problems, like the issue of weighting or the definition of thresholds.

Is a categorisation limited to conflict easier?

As noted earlier, the main challenge to traditional ways the aid industry does business is in conflict countries. Would it be easier to break the overaching concept of fragility back down to its basic elements, and focus on conflict?

To clear the decks, two useful definitions of conflict-affected and high risk areas are in Box 1, provided by the OECDOrganisation for Economic Cooperation and Development and the EUEuropean Union (and sourced from the Responsible Minerals Initiative). Note that the EUEuropean Union but not the OECDOrganisation for Economic Cooperation and Development references post-war fragility.

Box 1 " width="610" height="245" />Definitions of conflict-affected and high-risk areas

The World Bank includes a country on the list if (inter alia) there has been ‘the presence of a UN and/or regional peace-keeping or peace-building mission during the past three years’. 17 countries are identified as meeting this criterion: Afghanistan, Central African Republic, DRC, Cote d’Ivoire, Gambia, Guinea Bissau, Haiti, Kosovo, Liberia, Mali, Somalia, South Sudan, Sudan, West Bank/Gaza, Iraq, Lebanon and Libya. Several of those are not in the first two categories on the DACDevelopment Assistance Committee (of the OECD) list (for example, Liberia, Gambia, Lebanon).

If peace missions are an indicator, the EUEuropean Union currently has 16 ongoing military or civilian missions (Figure 4), including Moldova and Georgia: both are eligible for oda; neither is on the DACDevelopment Assistance Committee (of the OECD) high priority list.

Figure 4


The DIFD list of fragile states refers to the ‘World Peace Index’, presumably the Global Peace Index produced by the Institute for Economics and Peace. There are 16 countries scored red in the 2018 edition of the Index, including Ukraine and Russia (Figure 5). Ukraine is not on DFID’s list, though it is a country eligible for oda.

Figure 5

Global Peace Index 2018


The Global Peace Index is not the only option DFIDDepartment for International Development might have chosen. The Responsible Minerals Initiative has a list of resources and tools to help companies identify conflict-affected and high-risk areas. Useful resources listed include the Global Peace Index, and also: the Heidelberg Conflict Barometer; the Control Risks Worldmap; and the INFORM Worldmap. The States of Fragility 2016 Report mentions other sources: The George Mason University State Fragility Index; the Fragile States Index from the Fund for Peace; the Country Indicators for Foreign Policy (CIFP) fragility index; the Index of State Weakness in the Developing World; the Global Conflict Risk Index (by the way, an EUEuropean Union initiative); and the Peace and Conflict Instability Ledger. This begins to look like an industry. Note the enthusiasm for ranking and the construction of indices. Google ‘”fragile states” index’ and there are 250,000 hits. Check on Google Scholar and there are 16,500.

There is only limited overlap on these lists and indices also, as Sebastian Zaja and Javier Fabra Mata showed in their paper for the German Development Institute (albeit with data from 2008). The most important ‘controversial’ cases, which appeared on some lists but not others, included Cuba, Israel, Saudi Arabia, China and UAE.

The DAC, as noted, is best in class in our sample for the transparency of its classification. The key source is a series of Appendices in the 2016 States of Fragility Report. To cut a long story short, a list of indicators is subject to statistical interpretation using principal components in order to produce the classification. In the case of security, there are seven risk factors, potentially offset by five coping elements (Figure 6). Battle-related deaths, deaths by non-state actors and terrorism all feature on the list of risks, alongside criminal activity and the homicide rate. These can be offset by numbers of soldiers and police officers, among other things. When the principal components are calculated, the key indicators are rule of law and state control of territory, followed by armed conflict, terrorism, organised crime and interpersonal violence.

Figure 6

Security indicators from the DACDevelopment Assistance Committee (of the OECD) States of Fragility methodology


A relatively small number of countries on the DACDevelopment Assistance Committee (of the OECD) list are highly fragile for security reasons. They are: in the highest category, Somalia, South Sudan, Sudan, Yemen, Syria, Iraq, Afghanistan, and Libya (8 countries); in the second category, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Comoros, Congo (Brazzaville), DRC, Egypt, Eritrea, Ethiopia, Haiti, Iran, Kenya, Mali, Mauritania, Myanmar, Nepal, Niger, Nigeria, Pakistan, and Palestinian Territories (22 countries).

Finding a way forward

A few points.

First, there are A LOT of lists. It is important to be pragmatic. People draw up lists for different reasons, and donors, in particular, may benefit from a certain amount of constructive ambiguity. On the other hand, it is hard to have a sensible conversation when everyone has their own list. At the very least, people need to make their criteria explicit. Better, more harmonisation would be helpful, in order to have the kind of global conversation to which  Sebastian Zaja and Javier Fabra Mata point. They observe that indices, and by extension lists, have legitimate objectives. They

 ‘could be of use for development policy as a tool for:  determining which countries need a different aid approach; monitoring larger trends of global political stability; evaluating the overall impact of development aid; and for investigating the dynamics of state fragility.’

Second, however, the quest for quantitative rigour is very hard when data are lacking or out-of-date in many of the worst-affected countries. An alternative approach is to be less dependent on data, and to deploy it in the context of intelligence on the ground and expert knowledge. It is quite easy to imagine an exercise using the definitions in Box 1, perhaps using a panel of specialists to sift data and reach judgements about countries in conflict or at risk. James Surowiecki describes such an approach in The Wisdom of Crowds and it is being used by organisations like the World Economic Forum for the annual Global Risks Report. There are also echoes of approaches used in participatory development, giving a voice to people in affected countries.

Third, using qualitative data does not rule out constructing an index, if that is thought useful. The Transparency International Corruption Perceptions Index is an example of a global index constructed from qualitative data.

Fourth, a balance is needed between, on the one hand, recognising the complex inter-relationship between the different drivers of ‘fragility’, and, on the other, keeping the focus narrow enough to be useful. The risk of drawing the boundaries too wide can be seen in the OECDOrganisation for Economic Cooperation and Development States of Fragility Report 2018, the key messages of which include very general statements like ‘We will invest in more and smarter aid in fragile contexts’, ‘We will invest in the data to better understand, anticipate and respond to multiple states of fragility’, and ‘We will step up our efforts on prevention, peace and security’.

Furthermore, the universal approach risks losing sight of the underlying problems in conflict situations, which may be especially prominent in war zones, but which appear also in other places. These include, in addition to tackling the drivers of conflict: the interface between development and humanitarian actors and warring parties; the role of peace-making and peace-keeping forces; and the transition from war to peace. There is much policy-relevant work on all of those.



It turns out, then, that I am sympathetic to the argument that there are complex interactions between the different aspects of fragility, but sceptical about the conceptual and practical problems of building statistical models which will describe and, importantly, predict fragility.

It is possible to think of alternative approaches, using expert panels and participatory approaches, tailored to the needs of different actors. These should definitely be tried.

The focus, however, could usefully be redirected to the substantive issues which arise in conflict states, but also elsewhere.

Which raises the final question, of whether the term ‘fragile states’ has become debased. Too much has been loaded onto it. The term ‘conflict-affected state’ has longer legs, but also needs to defined with care and restricted to the egregious cases.

Appendix 1

Appendix 2


Appendix 3

DFID Fragile States Methodology


Appendix 4

Countries of operation of the UK Conflict, Security and Stability Fund

‘The CSSF is in over 70 countries including: Afghanistan, Algeria, Armenia, Azerbaijan, Bosnia-Herzegovina, Burma, Burundi, Colombia, Democratic Republic of the Congo, Egypt, Eritrea, Ethiopia, Georgia , Iraq, Jordan, Kenya, Kosovo, Kyrgyzstan, Lebanon, Libya, Macedonia, Mali, Morocco, Moldova, Nepal, Nigeria, Occupied Palestinian Territories, Pakistan , Peru, Serbia, Somalia, South Sudan, Sri Lanka, Sudan, Syria, Tunisia, Uzbekistan, Ukraine and Yemen. The CSSF also works in the UK Overseas Territories.’


Appendix 5



Note: a pdf version of this piece is here, with slightly more readable graphics.


# Fragile States - Part 1Joern.Graevingholt 2018-08-31 08:50
I have been struggling with the topic myself for quite some time and tried to make some contributions to the debate, including in discussions with the DAC’s International Network on Conflict and Fragility (aka INCAF). Most recently, our Constellations of State Fragility project went online at statefragility. info. (For background and some intro videos, see

I share your concern that in public discourse “fragile states” has tended to become a catch-all phrase with little analytic value. Lists of fragile states open up more questions than they claim to answer. Countries like North Korea and Mali, for instance, could hardly be more different, yet at times prominent indexes of state fragility ranked them in direct proximity to each other, leaving one wondering what to learn from such an exercise.

Definitions that are too broad (or hardly definitions at all but mere accumulations of characteristics ) have almost turned “fragile states” into another word for “countries with any kind of a serious problem”.

This is not where the debate started, though, and I believe there still is utility in using the terminology of state fragility. I am sure you noticed my shift from “fragile states” to “state fragility”, which is not accidental.
Reply | Reply with quote | Quote
# Fragile States - Part 2Joern Graevingholt 2018-08-31 08:51
Speaking of fragile states requires taking a decision on whether a state is fragile or not. In or out. Yes or no. But social and political realities rarely ever come in binary forms. Setting thresholds is therefore fraught with difficulties and most often simply misleading. Doing the same for a whole list of characteristics does not make the exercise more accurate, quite to the contrary.

Moreover, over time “fragile state” became synonymous with “fragile country” – another reason why the concept lost its analytical edge.

“State fragility”, by contrast, centres on a problem that can come in degrees and in different forms. The fragile states debate started in the mid-2000s on the heels of the “failing” or “failed” states terminology and the statebuilding paradigm. For all its political and ideological misuse in the wake of 9/11, one thing was remarkable about it: it was built on the recognition that countries and societies need a state and its institutions to govern them. And the quality of that state has tremendous consequences for a country’s development. (However, this should NOT lead one to believe that the state can or should be considered as the ONLY important determinant of development, peace or other valued goals. Nor that state institutions can be easily manipulated from outside.).

Analysing dimensions and patterns of state fragility as well as factors that contribute to their emergence can be extremely useful in order to understand current lock-ins and future risks. At the DIE, we use a three-dimension al concept of statehood that we have developed since 2012: state authority (monopoly of violence), state capacity and state legitimacy form the three core dimensions that determine the interaction between state and society. These dimensions can be measured and the patterns they form analysed – so that policy makers get a better sense of the TYPES of problems they encounter and researchers can investigate the causes and consequences of different types of fragility (we call them “constellations ”).
Reply | Reply with quote | Quote
# Fragile States - Part 3Joern Graevingholt 2018-08-31 08:52
With that in mind, here are some quick thoughts on your six conclusions:

1. Fully agree.
2. “Conflict” countries is itself a category fraught with difficulty. Is it only past and ongoing violent conflict? How about violence below a certain threshold? Or beside the media focus on armed rebel groups? How about the need of prevention (Guterres!) A focus on prior civil war would probably not have included Syria before 2011 – but arguably should have.

I guess my point is that there are more distinctions to have than conflict/non-co nflict or fragile/non-fra gile. This is what our Constellations try to offer.

3. Agree to the first sentence. And data quality is debatable in many instances. But even incomplete data can still help to structure reality in a meaningful way – as long as we remain aware of their limitations. Again, we believe our approach is one way of doing this while we are fully aware that for individual country analysis, macro data can not be more than a first step of an analysis that needs to get much deeper.

4. I agree for programming purposes at (target) country level. But policy planning starts earlier, doesn’t it? Here, (country) comparative frameworks are indispensable. If you do not introduce them explicitly, people will use their own implicit frameworks of comparison.

5. I do not agree to the extent that conflict is meant to replace fragility. (I mentioned above why I prefer state fragility over fragile states.) These are different foci that can overlap but need not do so.

6. I am not sure I understand what you mean by bottom up.
Reply | Reply with quote | Quote

Add comment

Security code
Security code:


latest pollVote now: 

Is the concept of 'fragile               states'                   over-                   burdened?


Follow me on Twitter